Academic journal article Journal of Money, Credit & Banking

Why Do Bank Runs Look like Panic? A New Explanation

Academic journal article Journal of Money, Credit & Banking

Why Do Bank Runs Look like Panic? A New Explanation

Article excerpt

EVEN THOUGH MOST empirical studies conclude that bank runs are associated with adverse information about banks (see Mishkin 1991, Hasan and Dwyer 1994, Saunders and Wilson 1996, Schumacher 2000, Calomiris and Mason 1997, 2003), the view that bank runs are caused by depositor panic is still popular because some features of bank runs cannot be fully explained by negative bank-specific information. For example, during large-scale bank runs, depositors usually do not distinguish between good and bad banks; they rush to all banks to withdraw deposits. An example is the 1932 Chicago Banking Crisis. (1) During the first week of that crisis (late June of 1932), massive withdrawals occurred from all banks in Chicago. As it is unlikely that all the banks in Chicago were financially distressed, it is fair to say that during this crisis some banks suffered runs even though they appeared sound. (2) Such runs may look more like depositor panic rather than the rational response of depositors to negative bank-specific information.

In this paper, we propose a new theory to explain why bank runs look like a panic. We define a panic run as a bank run that occurs when depositors' expectations of the bank's fundamentals do not change. In our model, panic runs are triggered by changes in depositors' expectations of the bank-specific information process. More specifically, depositors may start a run when they expect that more noisy information about banks will be revealed, or when they expect that precise information about banks will not be revealed. We show that panic runs can occur even if depositors are fully rational and always choose the Pareto-dominant equilibrium when there are multiple equilibria.

The intuition of our model can be explained as follows. Consider a bank that collects deposits to invest in risky assets. Suppose that depositors demand liquidity and the bank provides it by allowing early withdrawing depositors to consume more than the liquidation values of their deposits. Moreover, depositors may receive an interim signal about the return on the bank's assets.

In this setting, whether information-based bank runs can improve depositor welfare depends on the quality of the signal about bank assets. If the signal is relatively precise, information-based runs are beneficial because they can serve as a mechanism to efficiently liquidate the bank when its continuation value is lower than the liquidation value. On the other hand, if the signal is informative (3) but relatively noisy, then information-based runs will reduce depositor welfare. As shown in Chen and Hasan (2006), a deposit contract that provides liquidity will induce depositors to have excessive incentives to withdraw. The excessive incentives to withdraw may force depositors to respond to mildly adverse information about the bank and start a bank run even if they would be better off if the run did not happen.

The above results help explain why panic runs occur. At any point in time after they deposit, depositors can decide whether to withdraw immediately or to wait, and a bank run will occur when the depositors' expected payoff for waiting is lower than what they can receive from successfully withdrawing. When depositors learn that a relatively noisy (but still informative) signal will be revealed, they realize that a welfare decreasing bank run is more likely to occur, so their payoff for waiting becomes lower. Similarly, when depositors learn that a precise signal will not be revealed, they realize that they will not be able to use the signal for triggering a welfare improving bank run, so their payoff for waiting also becomes lower. In both cases, the reduction in depositors' expected payoff for waiting might lead to a panic run. In a more general sense, our results illustrate that information not only has the direct effect of changing posteriors, but also has the indirect effect of moving players along the game tree and affecting backward induction calculations. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.